Date: Mon, 29 Jul 2002 18:31:48 -0400 From: Wesley Eddy <weddy@masaka.cs.ohiou.edu> Subject: Re: tcptrace-bugs A few Questions Message-ID: <20020729183148.A4377@irg.cs.ohiou.edu>
On Mon, Jul 29, 2002 at 04:39:56PM -0400, Adam Plesniak wrote:
>
>
> Hello, my name is Adam Plesniak and I am doing some summer work with tcptrace
> and tcpdump and I had a few questions about the output of your program. At
> the bottom of a connection, the program lists the feilds < minowin:,
> maxowin:, averageowin:, and waverage owin:> What exaclty is an O window?
It's a kind of nonstandard term we use to describe the amount of 'o'utstanding
(unacked) data. There's a more detailed explanation in the manual at:
http://tcptrace.org/manual/owin.html
Hope that clears it up.
> And one more question, in reading recent papers aimed on predicting
> throughput of a connection mathematically, they refer to a Maximum Window
> variable which is constant during the duration of a transfer. From the
> tcptrace data, how can we derive this value to use in our own calculations.
> Any help you could give us would be greatly appreciated. Thanks for you
> time.
Well, it doesn't really have anything to do with a tcptrace-bug, but we
can probably answer this one too :). It's hard to say for sure without
any context, but I'd guess that they're refering to the receive buffer
size. Alternatively I guess it could refer to the total amount of data
the network could hold (sum of bandwidth delay product for all links in
the path as well as all queues in between) but I'm just guessing without
a specific source to look at and see what you're talking about.
-Wes
This archive was generated by hypermail 2b30 : 07/30/02 EDT